Simulation_dataset: separate in-house simulation and bppseqgen's

lib/simulation_dataset.mli

@@ -9,52 +9,18 @@ type tree =
       branch_length2 : float;
     }
 
-type 's benchmark = {
-  method_labels : string list ;
-  method_outputs : float option array list ;
-  average_precision : (float * (float * float)) list ;
-  site_model : 's array ;
-  ancestral_counts : int Phylogenetics.Amino_acid.table array ;
-  convergent_counts : int Phylogenetics.Amino_acid.table array ;
-}
-
 module type S = sig
-  type t
-  type site_model
+  type query
 
-  include Detection_pipeline.Query with type t := t
-  include Detection_pipeline.S with type query := t
+  include Detection_pipeline.S with type query := query
 
-  val alignment_plot : t -> svg file
-
-  val benchmark : t -> text file list -> site_model benchmark workflow
-
-  val rds_of_benchmark : site_model benchmark workflow -> rds file
+  val alignment_plot : query -> svg file
 end
 
-type t
-
-val bppseqgen_mixed :
-  ?ne_s:float ->
-  ?seed:int ->
-  tree:tree ->
-  profiles:string ->
-  n_h0:int ->
-  n_ha:int ->
-  unit ->
-  t
-
-val bppseqgen :
-  hyp:Convergence_hypothesis.t ->
-  tree:tree ->
-  profiles:string ->
-  nb_sites:int ->
-  seed:int ->
-  t
-
-module Mutsel_param : sig
-  type t
-  val make :
+module Mutsel : sig
+  include S
+
+  val query :
     ?branch_scale:float ->
     ?ne_s:float * float ->
     ?gBGC:float * float ->
@@ -64,62 +30,86 @@ module Mutsel_param : sig
     n_h0:int ->
     n_ha:int ->
     unit ->
-    t
+    query
+
   val simulation :
-    t ->
+    query ->
     Codepitk.Simulator.Site_independent_mutsel.simulation workflow
-end
 
-val mutsel :
-  ?branch_scale:float ->
-  ?ne_s:float * float ->
-  ?gBGC:float * float ->
-  ?seed:int ->
-  tree:tree ->
-  profiles:string ->
-  n_h0:int ->
-  n_ha:int ->
-  unit ->
-  t
 
-val of_mutsel_param : Mutsel_param.t -> t
+  type benchmark = {
+    method_labels : string list ;
+    method_outputs : float option array list ;
+    average_precision : (float * (float * float)) list ;
+    site_model : [`Convergent | `Non_convergent] array ;
+    ancestral_counts : int Phylogenetics.Amino_acid.table array ;
+    convergent_counts : int Phylogenetics.Amino_acid.table array ;
+  }
 
-include Detection_pipeline.Query with type t := t
 
-include Detection_pipeline.S with type query := t
+  val benchmark_statistics :
+    Codepitk.Simulator.Site_independent_mutsel.simulation workflow ->
+    labels:string list ->
+    results:(text file * int) list ->
+    binary_file file
 
-val oracle : t -> text file
+  (* val benchmark : t -> (t -> text file) list -> benchmark workflow
+   *
+   * val rds_of_benchmark : benchmark workflow -> rds file *)
 
-val alignment_plot : t -> svg file
+  (** stuff for gbgc SBME paper *)
+  type gBGC_t =
+    | Global of float
+    | Convergent of float * float
 
-val multinomial_benchmark : t -> pdf file
+  type param_t
 
-val benchmark : ?mode:[ `fast | `full ] -> t -> svg file
+  val explore_params : f:(param_t -> 'a) -> ((float * gBGC_t) * 'a) list
 
-val benchmark2 : t -> pdf file
+  val simu_of_param : ?n_h0:int -> param_t -> query
 
-val benchmark_statistics :
-  Codepitk.Simulator.Site_independent_mutsel.simulation workflow ->
-  labels:string list ->
-  results:(text file * int) list ->
-  binary_file file
+  val filter_results :
+    f:('a -> bool) -> (param_t * 'a) list -> (param_t * 'a) list
 
-(** stuff for gbgc SBME paper *)
-type gBGC_t =
-  | Global of float
-  | Convergent of float * float
+  type record_t
 
-type param_t
+  val record_of_simu : query -> record_t workflow
 
-val explore_params : f:(param_t -> 'a) -> ((float * gBGC_t) * 'a) list
+  val realistic_result : record_t -> bool
+end
 
-val simu_of_param : ?n_h0:int -> param_t -> t
+module Bppseqgen : sig
+  type t
 
-val filter_results :
-  f:('a -> bool) -> (param_t * 'a) list -> (param_t * 'a) list
+  val bppseqgen_mixed :
+    ?ne_s:float ->
+    ?seed:int ->
+    tree:tree ->
+    profiles:string ->
+    n_h0:int ->
+    n_ha:int ->
+    unit ->
+    t
+
+  val bppseqgen :
+    hyp:Convergence_hypothesis.t ->
+    tree:tree ->
+    profiles:string ->
+    nb_sites:int ->
+    seed:int ->
+    t
 
-type record_t
+  include Detection_pipeline.Query with type t := t
+
+  include Detection_pipeline.S with type query := t
+
+  val oracle : t -> text file
 
-val record_of_simu : t -> record_t workflow
+  val alignment_plot : t -> svg file
+
+  val multinomial_benchmark : t -> pdf file
 
-val realistic_result : record_t -> bool
+  val benchmark : ?mode:[ `fast | `full ] -> t -> svg file
+
+  val benchmark2 : t -> pdf file
+end